Using first-principles only, we calculate the melting point of MgO, also called periclase or magnesia. The random phase approximation (RPA) is used to include the exact exchange as well as local and non-local many-body correlation terms, in order to provide high accuracy. Using the free energy method, we obtain the melting temperature directly from the internal energies calculated with DFT. The free energy differences between the ensembles generated by the molecular dynamics simulations are calulated with thermodynamic integration or thermodynamic perturbation theory. The predicted melting temperature is T RPA m = 3043 ± 86 K, and the values obtained with the PBE and SCAN functionals are T PBE m = 2747 ± 59 K and T SCAN m = 3032 ± 53 K.

A density-functional-theory-based relativistic scattering formalism is used to study charge transport through thin Pt films with room-temperature lattice disorder. A Fuchs-Sondheimer specularity coefficient p ∼ 0.5 is needed to describe the suppression of the charge current at the surface even in the absence of surface roughness. The charge current drives a spin Hall current perpendicular to the surface. Analyzing the latter with a model that is universally used to interpret the spin Hall effect in thin films and layered materials, we are unable to recover values of the spin-flip diffusion length l sf and spin Hall angle sH that we obtain for bulk Pt using the same approximations. We trace this to the boundary conditions used and develop a generalized model that takes surface effects into account. A reduced value of sH at the surface is then found to describe the first-principles transport results extremely well. The in-plane spin Hall effect is substantially enhanced at the surface.